Rev. Biol. Trop. 50(1): 321-325, 2002 www.ucr.ac.cr www.ots.ac.cr www.ots.duke.

edu

BRIEF ARTICLE

A new method to assess air pollution using lichens as bioindicators

Julin Monge-Njera*1, Mara I. Gonzlez 2, Marta Rivas Rossi 1 and Vctor Hugo Mndez 1
1 2 * Centro de Investigacin Acadmica, UNED, San Jos, Costa Rica. Escuela de Estadstica, Universidad de Costa Rica, 2060 San Jos, Costa Rica. Reprint requests, mailing address: Biologa Tropical, Universidad de Costa Rica, 2060 San Jos, Costa Rica. Fax (506)2075550. Received 30-I-2001. Corrected 09-VII-2001. Accepted 09-VIII-2001.

Abstract: Lichens are increasingly used worldwide as air quality biomonitors because they are efficient, easy and cheap, but validation studies of the methodology are scarce. Three foliose lichen biomonitoring methods were compared by field tests (in the tropical urban habitat of San Jos, Costa Rica) and laboratory simulations: (1) the 100 uniform squares template traditionally used in North America, (2) the European 200 uniform points template and (3) a new computer-generated random points template (10 X 20 cm) in two versions: 100 points and 50 points. Repeated measurement by the same observer causes a variation of 2-14% and the templateserror is 0.2-11%. We recommend the 100 random point template (applied to four sides of trunk) for ecological studies and the 50 random points template (applied to side with greatest lichen cover) for biomonitoring because it reduces time and costs by nearly 50% but still has acceptable reliability values. Key words: Bioindication, lichens, air pollution, methodology.

Lichens are increasingly used as air quality biomonitors (Bartoli et al. 1997, Malysheva 1998) because they have several advantages over electronic monitors. These are expensive and their use and maintenance are not simple or cheap. Even industrialized nations are unable to afford extensive networks of electronic monitors. They are limited to a few elements or chemical compounds and have no intrinsic relationship with the biological effect of the contaminants (Rodrigo et al. 1999). In contrast, biomonitors are available for free and there are millions of them already functioning throughout the world (Ockenden et al. 1998). They integrally reflect the environmental influence over organisms and can be understood and used by the common citizen with minimal training. Even though almost all biological species can be considered biomonitors to some extent, the above advantages

apply particularly well to lichens when air pollution is considered (van Dobben and TerBraak 1999). In tropical regions, poor knowledge of lichen taxonomy does not affect basic biomonitoring because this method does not require species identifications (Carreras et al. 1998, Monge-Njera et al. 2001). Air biomonitoring is particularly developed in Europe (Faltynowicz 1997), where the lichen Hypogimnia physodes is used as a standard species (Grninger and Monge-Njera 1988) and much research is done about the involved biochemistry (Hyvarinen and Critenden 1998), but to our knowledge, no new simple field methods have been developed for at least three decades. This paper compares the reliability and cost of two traditional methods and describes a new computer-generated method that proved reliable and reduced costs by 50%. The phorophytes were selected because they

322

REVISTA DE BIOLOGATROPICAL

were available in representative parts of the city (Mndez and Fournier 1980), met the requirements for this type of study (Mndez and Fournier 1980), and because they had been under observation for two decades (MongeNjera et al. this issue).

MATERIALS AND METHODS Template types: The following methods were compared: (1) the 100 uniform squares (10X10 cm) template used in North America and applied to the four cardinal sides of tree trunks (Mndez and Fournier 1980), (2) the European 200 uniform points (10 x 20 cm) template applied to the tree trunk side that is visually identified as having the highest lichen cover (Grninger et al. 1980) and (3) a new random points template (10 x 20 cm) applied to the four cardinal sides of tree trunks. The random point positions were generated with standard random generator computer software and the 2 mm diameter points were applied with white paint on a translucent plastic sheet. The reliability of the three templates was examined under experimental conditions and, additionally, methods (2) and (3) were field tested in 1987 in the city of San Jos, Costa Rica and are compared here among themselves and with a parallel study with method (1) (Monge-Njera et al. 2001). For details about the reference site see Mndez and Fournier (1980). Reliability experiments: The differences produced by template design were measured by asking sixteen biology students to measure cover on the same section of a single trunk with the square, uniform point and random point templates. Real cover was measured by tracing the lichens on paper of homogeneous thickness and weighing the clippings to the nearest 0.0001 g. To assess the effect of repeated sampling, the same students estimated lichen cover by blindly placing a template ten times over a drawing of bark with one dominant and one scarce lichen species (N = 160 measurements,

Fig. 1. Drawing of bark with one dominant and one scarce lichen species used to test template effect.

Fig. 1). Real cover was measured by clipping the drawings on paper of homogeneous thickness and weighing to the nearest 0.0001 g. This experiment was done with the uniform point template and with the 100 random point template. Additionally, the effect of sample size was examined in the point templates by calculating twice: with all data and with a random subsample of half the data (i.e. 50 points).

RESULTS Field test: In one half of the stations, the uniform point method (Table 1) produced sig-

INTERNATIONALJOURNALOF TROPICALBIOLOGYAND CONSERVA TION

323

TABLE 1 Comparison of uniform point and the random point method results in the stations (San Jos, Costa Rica, 1987) and F test significance values* Station number Uniform point method Random point method, mean values 1.10 3.50 4.60 0.10 19.80 9.50 8.70 18.20 0.60 43.52 F significance Random point method, maximum values 2.9 8.2 12.0 0.5 36.8 23.5 33.5 51.4 2.6 70.0 F significance

2 3 4 5 6 7 8 9 10 11

8.70 15.00 6.50 1.90 22.70 12.90 29.60 47.50 1.50 17.75

0.0005 0.0366 0.8245 0.0000 0.8010 0.5264 0.0025 0.0837 0.0388 0.0028

0.1546 0.8744 0.0122 0.0019 0.0903 0.1686 0.5450 0.2791 0.0664 0.0004

* These methods were applied to measure lichen cover in the city of San Jos, Costa Rica in 1987 as part of a long term study and the results are published somewhere else (Monge-Njera et al. 2001).The uniform point method always uses the tree side with highest lichen cover; the random point method used the four tree sides in this case only for comparison: it is fit for use on the side with the highest cover. For comparison, the random point method used both the highest and the mean values for each tree. There were no lichens in Station 1.

nificantly higher estimates than the random points method applied to the four sides. When both methods were applied to the side with the highest cover, they produced statistically equivalent results. Reliability experiments: Repeated measurements by the same observer causes a variation range (standard deviation) of 2-14% (Table 2). Relative lichen cover estimated when several observers measured the same tree with three templates were: 100 squares method: real cover 6%, estimated cover 17% (17; 1-79), N = 22. 200 uniform points method: real 51%, estimated 44% (17; 22-82), N = 21. 100 random points method: real 51%, estimated 59% (18; 30-95), N = 22. The squares method overestimated by 11%. The uniform points method underestimated by 7%, and the random points method overestimated by 8%. In the laboratory simulation, results differed significantly among templates (Kruskal-

Wallis ANOVA and Tukey Test, p < 0.01) but errors are not large (0.2-11%). In comparison with using all the points, templates of the same size and shape but with only half the points are only about 5% less reliable (range 0.3-6.9%) than the full versions (Table 2).

DISCUSSION The spatial microdistribution of lichens is associated with atmospheric pollutants in a complex way because pollutant impact depends on substrate texture, shape and inclination, wind, light, humidity, organism activity and other factors (van Dobben and TerBraak 1999). Furthermore, the association is highly variable (e.g. Monge-Njera et al.); for this reason we recommend the 100 random point template (applied to four sides of trunk) for ecological studies, which requires a precise description of spatial distribution. The physiology of the lichen-environment interaction was studied among others by Malhotra and Khan (1983). When lichens are used as bioindicators for air pollution monitoring (Carreras et al. 1998, Hyvarinen and Crittenden 1998), we

324

REVISTA DE BIOLOGATROPICAL

TABLE 2 The effect of repeated measurement with two template types and two sample sizes on estimation of cover by a dominant and a minoritary lichen species (laboratory simulation). Each row: N = 160 Lichen cover (%) estimated by each method Dominant species Random template Subsample Total sample Uniform template Subsample Total sample Real value Minoritary species Random template Subsample Total sample Uniform template Subsample Total sample Real value Mean Standard deviation Minimum Maximum

55.4 51.4 40.4 52.3 51.2

12.4 11.4 13.6 1.9

20 25 15 48

90 70 59 58

16.9 12.4 9.7 11.5 10.0

12.7 6.5 3.6 3.1

0 0 2 5

110 30 20 25

recommend the 50 random points template (applied to the side with greatest lichen cover) because in comparison with the other methods it reduces time and costs by nearly 50% but still has acceptable reliability values (4.2% error, 12.4% standard deviation).

ACKNOWLEDGMENTS We thank Werner Grninger for academic advise as well as Harrie Sipman (Berlin Herbarium), Luis A. Fournier (University of Costa Rica) and Artemis Papert (Geneva), whose comments on earlier drafts allowed us to greatly improve the manuscript. Our thanks also to the University of Costa Rica biology students who assisted with the experiments and to Marisol Rodrguez Pacheco for the illustration.

100 cuadrados uniformes usada en Amrica del Norte, (2) la plantilla europea de 200 puntos distribuidos uniformemente y (3) una nueva plantilla generada por computadora que consta de puntos aleatoriamente ubicados en un rectngulo de 10 X 20 cm, en dos versiones: 100 puntos y 50 puntos. Los errores debidos al observador o a la plantilla son de 0.2-14%. Para estudios ecolgicos se recomienda por su precisin la plantilla de 100 puntos aleatorios aplicada a los cuatro puntos cardinales del tronco y para monitoreo de contaminacin, la de 50 puntos aleatorios aplicada al lado con mayor cobertura porque produce resultados semejantes a los mtodos tradicionales pero reduce tiempo y costos a la mitad.

REFERENCES
Bartoli, A., C. Cardarelli, M. Achilli, L. Campanella, S. Ravera & G. Massari. 1997. Quality assessment of the Maremma Laziale area using epiphytic lichens. Allionia 35: 69-85. Carreras, H.A., G.L. Gudino & M.L. Pignata. 1998. Comparative biomonitoring of atmospheric quality in five zones of Crdoba city (Argentina) employing the transplanted lichen Usnea sp. Envir. Pollut. 103: 317-325. Faltynowicz, W. 1997. Lichens as indicators of bog-community degeneration. Acta Mycol. 32: 347-368.

RESUMEN
Se compar tres mtodos para la medicin de cobertura de lquenes en el campo (ciudad de San Jos, Costa Rica) y en simulaciones de laboratorio: (1) la plantilla de

INTERNATIONALJOURNALOF TROPICALBIOLOGYAND CONSERVA TION

325

Grninger, W., E. Gruner & I. Schnitzer. 1980. Flechten und luft qulitat. Luftqualitt in Freirumen und Verdichtungsrumen, angezeigt durch Rindenflechten (IAP-Verfahren). Stadtverwaltung, Reutlingen, Germany. Grninger,W. & J.Monge-Njera. 1988. Use of the temperate lichen Hypogymnia physodes (Parmeliaceae) to evaluate air pollution in the tropics. Rev. Biol. Trop. 36: 545-547. Hyvarinen, M. & P.D. Crittenden. 1998. Relationships between atmospheric nitrogen inputs and the vertical nitrogen and phosphorus concentration gradients in the lichen Cladonia portentosa. New Phytol. 140: 519-530. Malhotra, S.S. & A.A. Khan. 1983. Sensitivity to SO2 of various metabolic processes in an epiphytic lichen, Evernia mesomorpha. Biochem. Physiol. Pflanzen 178: 121-130. Malysheva, N.V. 1998. Lichens of St. Petersburg. 3. The influence of town environment and lichen indication of atmospheric pollution. Bot. Zh. St. Petersburg 83: 39-45.

Mndez, O. & L.A. Fournier. 1980. Los lquenes como indicadores de la contaminacin atmosfrica en el rea metropolitana de San Jos, Costa Rica. Rev. Biol. Trop. 28: 31-38. Monge-Njera, J., M.I. Gonzlez, M. Rivas R. & V.H. Mndez-Estrada. 2001. Twenty years of lichen cover change in a tropical habitat (Costa Rica) and its relation with air pollution. Rev. Biol. Trop. 50: 309-319. Ockenden, W.A., E. Steinnes, C. Parker & K.C. Jones. 1998. Observations on persistent organic pollutants: Implications for their use as passive air samplers and for POPcycling. Envir. Sci. Technol. 32: 2721-2726. Rodrigo, A., A. vila & A. Gmez-Bolea. 1999. Trace metal contents in Parmelia caperata (L.) Ach. Compared to bulk deposition, throughfall and leafwash fluxes in two holm oak forests in Montseny (NE Spain). Atmos. Envir. 33: 359-367. Van Doven, H.F. & C.J.F. Ter-Braak. 1999. Ranking of epiphytic lichen sensitivity to air pollution using survey data: A comparison of indicator scales. Lichenologist 31: 27-39.

Editor: Jos A. Vargas